Spatial models of vote choice predict that valence factors like candidate quality matter less to voters as differences between policy platforms increase. However, this and related claims are hard to test because it is difficult to measure candidate quality. I construct a measure of candidate quality differences using over 23,000 endorsements from newspapers around the United States. I estimate candidate quality differences in elections between 1950-2020 to evaluate the effects of quality on election outcomes. I find that the higher quality candidate wins in a large majority of elections in the United States. A one-standard-deviation increase in relative quality increases two-party vote share by 4 percentage points. Contrary to popular beliefs, the effect of quality differences on vote shares has actually increased slightly over time. However, the decrease in competitive elections over the same period has reduced the share of elections where candidate quality can plausibly alter the election’s outcome.